HAL Id: ijn_00000563
https://jeannicod.ccsd.cnrs.fr/ijn_00000563
Submitted on 25 Jan 2005
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Scott Atran, Douglas Medin, Norbert Ross
To cite this version:
Scott Atran, Douglas Medin, Norbert Ross. The Cultural Mind: Environmental Decision Making and Cultural Modeling Within and Across Populations. Psychological Review, American Psychological Association, 2005, 112. �ijn_00000563�
To appear in Psychological Review (2005)
The Cultural Mind:
Environmental Decision Making and Cultural Modeling Within and Across Populations
Scott Atran
Centre National de la Recherche Scientifique, Paris The University of Michigan, Ann Arbor MI
(satran@umich.edu)
Douglas L. Medin and Norbert O. Ross
Northwestern University, Evanston IL Vanderbilt University, Nashville, TN (medin@northwestern.edu) norbert.o.ross@vanderbilt.edu
Abstract. This paper describes a cross-cultural research project on the relation between how people conceptualize nature (their mental models) and how they act in it. Mental models of nature differ dramatically among and within populations living in the same area and engaged in more or less the same activities. This has novel implications for environmental decision making and management, including dealing with commons problems. Our research also offers a distinct perspective on models of culture, and a unified approach to the study of culture and cognition.
We argue that cultural transmission and formation does not consist primarily in shared rules or norms, but in complex distributions of causally-connected representations across minds in interaction with the environment. The cultural stability and diversity of these representations often derives from rich, biologically-prepared mental mechanisms that limit variation to readily transmissible psychological forms. This framework addresses a series of methodological issues, such as the limitations of conceiving culture to be a well-defined system or bounded entity, an independent variable, or an internalized component of minds.
I. Introduction.
This paper describes an ongoing body of research on environmental decision making.
Our research program is framed in terms of folk biology and the focus is on the relationship between how people conceptualize nature and how they act in it. The context of primary interest is the lowland rainforest of Guatemala, but we will also bring evidence to bear from other Mesoamerican and North American settings where resource conservation is at issue.
Perhaps the most central problem in environmental decision making is Hardin’s (1968)
“tragedy of the commons,” where individuals acting according to their self-interest overuse and deplete resources. But this gloomy outcome does not always come to pass. A number of
researchers have noted many examples where commons have been, and are being, successfully managed (Atran, 1986; Berkes, et al, 1989; Ostrom, 1999; Deitz et al., 2003). Key factors in success include a closed-access system and having social institutions in place to monitor use and punish cheaters (overusers, free riders).
Although it is comforting to know that the tragedy of the commons is not inevitable, current conditions of globalization do not conform well with the constraints that have so far been identified. For a variety of reasons, closed access is increasingly difficult to achieve, and local institutions are increasingly confronted by, and giving way to, inter-group conflict over resource use. Our research program provides a new theoretical perspective on resource dilemmas,
particularly those involving multiple cultural groups. We argue that how people conceptualize nature is linked with how people act in relation to it. In addition, we believe that cultural
differences in mental models and associated values play an important role in creating inter-group conflict and, therefore, may hold the key to addressing these conflicts.
At the level of method, we offer techniques that are designed to facilitate cross-cultural research and comparison, but also aimed at avoiding the pitfall of reifying “culture” and treating it as an independent variable. Reifying culture wrongly equates reliable results with cultural patterning and homogeneity.
Our research program is also closely tied to a distinct stance with respect to the conceptualization of culture and cultural models. Just as Darwin had to rely on an intuitive notion of species in the processes of ultimately deconstructing it, our approach is not aimed at establishing some true definition or identification of “culture” but rather to understand the factors and dynamics that lead to cultures as constructed categories. We begin with “cultures” as
commonly described units - self-identified groups of individuals - and then trace the distribution of patterns of knowledge and beliefs both across and within cultures. It is these distributions, including their development and change that constitute the object of our study.1
From this standpoint, it is less useful to try to estimate population parameters for cultural norms – that is, shared values, beliefs and associated behaviors – than it is to establish the pathways that determine how (in our case, biological) ideas affect (in our case, environmental) behaviors (and vice versa). Accordingly, we describe emergent agreement patterns that are derived statistically from measurements of inter-informant agreement. To the extent that these agreement patterns overlap with patterns of self-ascribed cultures, we refer to such emerging models as “cultural models.” The focus of the research, however, is not a search for agreement patterns that overlap with self-ascribed cultural units, but rather understanding the underlying causes and dynamics of emerging agreement patterns.
In the present paper, our focus is on illustrating in detail how the distributional view of culture plays a critical role in our program of research on environmental decision making, although the utility and implications of this approach extend beyond our specific examples.
The rest of this paper is organized as follows. First, we review strengths and limitations of alternative views of culture. We go on to present a distributional view of culture and
methodological tools tailored to it. Then we briefly describe the settings and populations for our research studies. With this as background, we proceed to review our research findings on mental models of the environment and their correlations with behavior.
II. Culture as a Notional, not Natural, Kind
Intuitively, one might define culture as the shared knowledge, values, beliefs and practices among a group of people living in geographical proximity who share a history, a language, and cultural identification (see Brumann, 1999 and associated commentaries for examples of this approach in anthropology). From a psychological perspective, Campbell’s (1958) proposed measures of social entitativity in terms of common fate, similarity, proximity, resistance to intrusions and internal diffusion seem applicable to cultural groups.
But it is important to note that question of how culture should be defined is separable from the question of how best to study it. Although we think a definition of a culture in terms of history, proximity, language and identification is useful and (if not too rigidly applied) perhaps even necessary as a beginning point, it does not follow that the cultural content of interest must be shared ideas and beliefs.
It’s not easy to escape from this intuitive notion of culture any more than it is easy for biology to escape from the notion of species as ahistorical, well-bounded entities sharing an underlying essence (e.g. Mayr, 1989). In the same way that cultures are not natural kinds,
biological kinds don’t have the stable characteristics often attributed to them. Modern
evolutionary biology is the study of change and not just stability. Continuing this parallel with evolutionary biology, we believe that modern cultural research must be able to overcome intuitive notions of culture in order to focus on causal processes associated with stability and change. Both biological and cultural research started with folk notions (of species and culture, respectively) and they have served each field well as starting points. Ultimately, though, such conceptions must be radically altered for further progress to be made.
In the General Discussion we will return to the parallels between species in biology and intuitive notions of culture in cognitive science. Bearing in mind these issues concerning stability and change, we turn now to current stances on how culture and cultural processes should be studied. Each of them is useful for some purposes and all of them have limitations.
1. Culture as Norms and Rules. It appears natural to think that the cultural contents of interest must be shared in order to qualify as “cultural.” Note, however, that this commitment undercuts the dynamic side of cultural processes: distinctive values, beliefs and knowledge might or might not be consensual within a culture. For example, a culture may have a set of beliefs and practices known only to a privileged group of people (e.g. healers, elders, ruling elite) that
nonetheless are powerful forces within a given culture (and distinguish one culture from
another). In short, this view of culture as shared beliefs and practices not only prejudges the issue of what constitutes cultural content, but also, as a consequence, directs attention away from understanding the dynamic nature of social processes.
Some influential models of culture formation and evolution in biology and anthropology take a somewhat more liberal view of consensus. They are based on group-level traits that assume cultures are integrated systems consisting of widely-shared social “norms” (“rules,”
“theories,” “grammars,” “codes,” “systems,” “models,” “worldviews,” etc.) that maintain heritable variation (Rappaport, 1999; Laland et al., 2000; Wilson, 2002). Some political scientists also tend to view cultures as socially “inherited habits” (Fukuyama, 1995), that is, as socially transmitted bundles of normative traits (Huntington, 1996; Axelrod, 1997a).
The interest in heritable variation loosens the restrictions on consensus and raises
questions about the basis for variation. But here cognitive scientists are likely to be disappointed by the implicit assumption that the gist of cultural learning is the (more or less automatic)
absorption of norms and values from the surrounding culture (by processes no more complicated than imitation). We believe that there are two problems with such an approach: First, it is not clear how people would decide what exactly to imitate. Second, these assumptions do not pay sufficient attention to the sorts of inferential and developmental processes that allow human beings to build and participate in cultural life.
2. Cultural Psychology. The recent upsurge of interest in cultural psychology (for one review and critique, see Oyserman, Coon & Kemmelmeier, 2002 and associated commentaries) has produced a variety of intriguing findings and has done psychology a service by calling attention to cultural variation. Many of these studies show that knowledge systems previously thought to be universal actually vary widely across the world (for a review, see Cohen, 2001).
The lesson drawn is that: “Psychologists who choose not to do cross-cultural psychology may have chosen to be ethnographers instead” (Nisbett, et al., 2001:307). In brief, cultural psychology is succeeding in divesting academic psychology of implicit and ingrained ethnocentric biases.
What defines or constitutes cultural psychology? The area draws much of its inspiration from researchers such as Hofsteder (1980) and Triandis (1995) who sought to characterize cultural differences in terms of a small number of relevant dimensions. The project is successful
if multiple sources of evidence converge on the same small set of dimensions. Examples of such dimensions that have received a lot of attention are individualism versus collectivism and
egalitarian versus hierarchical social structure. Other researchers such as Nisbettt (2003) have used socio-historical analysis to derive dimensions of cultural differences in worldviews or preferred modes of thought. Examples of these dimensions are analytic and logical (categorical, axiomatic and non-contradictory) versus holistic and dialectical (thematic, no first principles or excluded middle). In short, Nisbett and his associates are suggesting that cultural studies must include not only contents per se, but also thinking processes that themselves may be
differentially distributed across cultures.
Cultural psychologists import the rigor and controls of standard experimental procedure into anthropological concerns, providing clear identification of the participants, thoughts and behaviors tested. Cultural psychologists are thus able to systematically exploit anthropological insights to demonstrate that mainstream psychology’s long-held assumptions about cognitive processes can be quite mistaken.
In our opinion, cultural psychology has some limitations. The leap from statistical regularity in some sample population to “the culture” may suffer from precisely the sort of reasoning criticized in mainstream psychology’s leap from Americans or Europeans (or, more typically, psychology undergraduates) to the world at large. The same inchoate conception of culture once used by many anthropologists and still used by most ordinary folk remains
customary in much cultural psychology. In this view, culture becomes a stable and shared set of beliefs, practices or strategies to be studied as yet another population parameter / personal attribute. This ahistorical, consensual view of culture limits the ability to explain and understand cultural differences once they are encountered. In other words, it is not clear how explanation or
interpretation can be extended beyond simple description. In some cases researchers have been able to exert some experimental control by priming tendencies to act individualistically versus collectively (e.g. Gardner, Gabriel and Lee, 1999, Briley, Morris and Simonson, 2000). These sorts of studies reinforce the dimensional analysis and potentially extend its scope. There is always the risk, however, of circularity in analysis. If priming does not affect some candidate task measuring individualism versus collectivism, then maybe the prime was ineffective or the task does not entail individualism and collectivism.
Perhaps we are guilty of prejudging the initial phase of a two-stage project. In stage one cultural psychologists tend to characterize culture as an external, historically-determined system that becomes internalized in the individual through “acculturation” (or some other causally opaque process), either diffusely or as some specialized part of the psyche responsible for
cultural (or social) cognition. A stage two focus on within-culture variations in modes of thought might illuminate how different cultural institutions shape ways of thinking and vice versa.
For cultural psychologists trained as anthropologists, the focus is on the “extrasomatic”
or “extragenetic” nature of culture as an integrated corpus of external control mechanisms that program individual minds and bodies, molding them in patterned ways recognizable across individuals (Geertz, 1973). We agree that expressions of the human psyche are profoundly embedded and structured within social and historical contexts, but we dissent from the invited implication of a one-way influence, with individual minds being passive recipients of “culture.”
So far we have followed current practice in using the term, cultural psychology, to describe the recent upsurge of cross-cultural comparisons by cognitive and social psychologists.
This may be a bit misleading in that one of the pioneers of the use of the term, Richard Shweder,
uses it refer to a set of ideas that entail rejecting psychic unity as well as rejecting the idea of characterizing cultural differences as variation along a small number of dimensions:
“[C]ultural psychology interprets statements about regularities observed in a lab or observed anywhere else, on the street or in a classroom, in Chicago or in Khartoum, not as propositions about inherent properties of a central processing mechanism for human psychological functioning, but rather as descriptions of local response patterns contingent on context, resources, instructional sets, authority relations, framing devices, and modes of construal.” (Shweder, 1991, p.87)
To avoid confusion in nomenclature, we will categorize Shweder’s approach to cultural psychology under the next framework, context and situated cognition.
3. Context and Situated Cognition. There are alternative views of “cultural psychology”
that call into question the use of standard forms of experimental procedure (“methodological behaviorism”) as fundamentally flawed on grounds that they are ethnocentrically biased in their focus on the individual mind/brain. Instead of considering cognitions to be embedded
exclusively in individual minds - with “culture” as just one component of individual cognition - these theorists maintain that human cognitions should be properly situated in cultural-historical context and “practical activity” (Cole 1996; cf. Vygotsky, 1978). A related concern is that cultural cognitions may be better understood as “distributed cognitions” that cannot be described exclusively in terms of individual thought processes, but only as “emergent structures” that arise from irreducible levels of interactional complexity involving differential linking of individual minds in a given population (Hutchins, 1995).
Researchers such as Michael Cole believe that culture cannot be entirely conceptualized in terms of cognitions, belief systems and the like, but must instead consider a culture’s artifacts
(construed broadly enough to include language). Cole (1996) argues that subjects and objects are not only directly connected but also indirectly connected through a medium constituted of artifacts. These artifacts are simultaneously material and conceptual. One consequence of this view is an emphasis on studying cognition in context where cognitive labor may be distributed across individuals as well as artifacts (such as plumb lines or computers). Since context includes people’s conceptions of artifacts, it is inherently relational.
We share some of these concerns raised by the situated view, such as (1) difficulties with standard experimental procedures, including 2 X 2 designs with culture, in effect, treated as an independent variable (Medin & Atran, 2004), and (2) lack of concern with differential
distributions of cognitions among minds within populations. For example, with respect to shared knowledge and beliefs Cole says: “in order to say anything useful, it is necessary to specify sources of coherence and patterning as a part of the ongoing activities that the inquirer wants to analyze” (Cole, 1996:124). We also agree that a focus on norms and rules is overly narrow, that cultural notions are intimately tied to the study of development, and that one good research avenue involves looking at how cognition plays out in particular contexts.
Other aspects of the situationist view seem vague. The idea that cognition is “stretched across mind, body, activity, and setting” is a useful framework notion that leads one to consider more than individual minds. At the same time, however, we believe that cultural situations and institutions cannot literally enter individual minds; rather, like other sorts of environmental stimuli they stimulate (in controlled and sequenced ways) mental processes that construct representations in accordance with a host of internal constraints, including evolved cognitive aptitudes like the folkbiology module (Medin & Atran, 2004). Cole (1996, p. 198) agrees with this assessment of internal constraints: “According to the version of cultural historical
psychology I am advocating, modularity and cultural context contribute jointly to the development of mind.”
Perhaps a fair summary is that claims about cultural, historical analyses represent
something of a promissory note (with respect to individual cognition) and research has tended to focus on situations and practices rather than the mediating mental representations associated with them. Strategically, this makes a certain amount of sense. In commenting on this section of this paper Ed Hutchins (Personal Communication, 11/9/2004) said: “If we situated guys have erred on the side of focusing on ‘situations and practices rather than the mediating mental
representations associated with them’ it is because the latter have received plenty of attention, and the former are so understudied that their role in constituting the human mind has not been appreciated or understood by the majority of cognitive scientists. Furthermore, I believe that a better understanding of the former will change what we think to be accomplished by the latter.”
Our only disagreement with this is that although cognition has been extensively studied,
cognition in context has not; hence we see a continuing need to attend to mental representations.
4. Culture as a Superorganism. One of the oldest, and most persistent, approaches to “the science of culture” is to consider culture an ontologically distinct “superorganism” whose “laws”
are sui generis and do not arise from individual thoughts and behaviors, but which govern how individuals think and behave in social contexts (White, 1949). Anthropologist A.L. Kroeber (1923) first formulated the doctrine in this way:
“Culture is both superindividual and superorganic…. [T]here are certain properties of culture – such as transmissibility, high variability, cumulativeness, value standards, influence on individuals – which is difficult to explain, or to see much significance in, strictly in terms of organic personalities and individuals.”
This “American” school of cultural anthropology, which viewed culture as a superorganism, soon merged with the “British” school of social anthropology known as
“functionalism” (Evans-Pritchard, 1940). Functionalism holds that the beliefs, behaviors and institutions of a society function with machine-like regularity of a well-adapted organism so as to promote the healthy functioning of social groups. According to A.R. Radcliffe Brown (1950):
“In reference to any feature of a system we can ask how it contributes to the working of the system. That is what is meant by…its social function. When we succeed in
discovering the function of a particular custom; i.e., the part it plays in the working system to which it belongs, we reach an understanding and explanation of it.”
For the last half century, anthropology has mostly abandoned pretensions to a “science of culture” based in the law-abiding functional regularity of the adaptive superorganism.2 But this view has recently made a comeback under the evolutionary guise of “group selection.”
According to philosopher Elliot Sober and anthropologist David Sloan Wilson (Sober & Wilson 1998:150-176): “In most human social groups, cultural transmission is guided by a set of norms that identifies what counts as acceptable behavior,” and which “function largely (although not entirely) to make human groups function as adaptive units.” Norms are functioning parts of a
“complex and sophisticated machine designed to forge groups into corporate units.”
From this level of analysis, mental structures can be effectively ignored when trying to make scientific sense of culture. Although human cultures perhaps developed “to function as adaptive units via many proximate mechanisms” (Sober & Wilson, 1998:182), it is possible to study cultures as “phenotypes” without describing the proximate computational machinery that generates them:
As long as the proximate mechanisms result in heritable variation, adaptations will evolve by natural selection. There is a sense in which the proximate mechanism doesn’t matter. If we select for long wings in fruit flies and get long wings; who cares about the specific developmental pathway?… if humans have evolved to coalesce into functionally organized groups, who cares how they think and feel? (Sober & Wilson, 1998:193; see also Dennett, 1995:358-359).3
We believe, however, that understanding cultural formation and evolution depends profoundly on understanding the “proximate” cognitive mechanisms involved. Perhaps we can best summarize with an analogy: macroeconomics is a legitimate field of study and generates important insights into economic activity on the basis of assumptions, for example, of an efficient market (and optimal individual behavior). But these insights do not in the least
undermine microeconomics and, more to the point, observations from microeconomics, such as loss aversion (e.g. Kahneman and Tversky, 1979) and mental accounting (e.g. Thaler, 1985), have had a significant impact on macroeconomics.
5. The Grammar of Culture. In anthropology, there is a long tradition of considering culture along the lines of language, that is, as being a rule-bound system with its own
“grammar.” This view of culture is most strongly associated with the “structuralist” school of Claude Lévi-Strauss (1963) in France and Mary Douglas (1970) and Edmund Leach (1976) in Great Britain. On this account the bewildering variety of social phenomena and cultural productions are variations generated from a universal structure of the mind (a grammar of culture), which allows people to make sense of the world by superimposing a structure based on a few underlying principles. The structuralist’s task is to gather as many variations as possible of some grammatical subsystem of culture (e.g., myth, kinship) in order to identify the most
fundamentally meaningful components in the sub-system, and to discern the structure through the observation of patterning. Following the linguistic theory of Ferdinand Saussure in which phonemes (the smallest unit of linguistic meaning) are understood in contrast to other phonemes, structural anthropologists argued that the fundamental patterns of human thought are also based on a system of “binary contrasts” to produce more elaborate systems of cultural meaning.4
Structural anthropology had little knowledge of the theories of cognitive architecture developed over the last few decades by cognitive and developmental psychologists,
neuropsychologists or generative linguists. The fundamental properties attributed to the human mind, such as “binary contrast,” were few and simple-minded (or so general and vague as to be applicable willy-nilly to any phenomena at all).5 This is not to deny the insights that structural anthropologists garnered into the relationships between different aspects of cultural life within and across populations (e.g., linking myth, kinship, folkbiology, hunting and cooking practices, residential architecture, etc.). Instead, it is only to deny that structuralist theories provide any principled causal explanation concerning how these relationships might have come about.
More current anthropological views of the grammar of culture are less committed to a specific theory of the cognitive architecture responsible for cultural productions than to the belief that culture consists of a bounded set of rule-bound systems, each with its own grammar-like structure. A more recent work in linguistic anthropology describes the “culture as grammar”
view as follows:
to be part of a culture means to share the propositional knowledge and the rules of inference necessary to understand whether certain propositions are true (given certain premises). To the propositional knowledge, one might add the procedural knowledge to carry out tasks such as cooking, weaving, farming, fishing, giving a formal speech,
answering the phone, asking for a favor, writing a letter for a job application. (Duranti, 1997:28-29)
Anthropology, then, is the discipline of “writing” the grammar of culture (Keesing, 1972:302). From this perspective, it seems that virtually any patterned activity that numbers of people share in can be considered “grammatical,” from pottery making to story telling. For example, “religion belongs to the elementary grammar of culture” (Kannengeiser 1995; cf.
Lawson & McCauley 1990). But there may be nothing interestingly “grammatical” (generated by few and finite rules) about how various cognitive systems link up together to make up “religion”
(Atran, 2002; Atran & Norenzayan, in press) or “science” (Atran 1990; Atran 1998) or “culture.”
6. I-Culture. A somewhat similar view, that is more sophisticated but also problematic, has recently arisen among (some) evolutionary psychologists. It is modeled on Noam Chomsky’s distinction between the internal, individual grammar that a given person possesses (“I-
Language,” such as someone’s particular knowledge of American English) and the external language (“E-Language,” such as the countless dialects, words, and stylistic differences of the English language as it has developed across the world over the last thousand years or so). Just as the English language was shaped – and is still being shaped - by broad historical events that did not take place inside a single head (including the Norman invasion of England and the global internet), so too did Western European or Chinese or Navajo culture (Pinker 1992:71).6
If the analogy holds, then psychology’s contribution to understanding “culture” might best focus on how children "grow" an I-Culture through the combination of an innate,
biologically specified "culture acquisition device" and the exposure to stimuli in the world (or, equivalently, how individuals are capable at all of participating in “E-Culture”). As Gary Marcus (2004:27) proposes: “The very ability to acquire culture is, I would suggest, one of the mind’s
most powerful learning mechanisms.” This suggests a line of inquiry for culture studies parallel to that taken by generative linguistics over the past fifty years, in which the fundamental guiding questions include: "What do people know when they know 'culture'?" and "How do people come to acquire 'culture'?”
Unlike the structuralist version of “culture as grammar,” this version does not prejudge the complexity or variety of cognitive mechanisms that may be involved in cultural acquisition.
Like more current anthropological versions, however, it seems to assume that I-Culture is a bounded system, or an integrated collection of systems, generated (under appropriate experience) by some articulated set of cognitive principles.
But we contend that there is no systematically bounded or integrated culture as such.
There is nothing at all “grammatical” or generatively rule-bound about the relations that connect, say, language, religion, the nation-state and science (or that connect the capacities to acquire knowledge of, and participate in, languages, religions, nation-states, and sciences). There are only family resemblances to what is commonsensically referred to as “culture” (or “religion” or
“science”), but no overarching or integrated structure.
7. Generativist (Agent-Based) Models of Culture. Recent advocates of agent-based computational models of cultural phenomena also sometimes borrow self-consciously from the framework of generative linguistics, where few and finite rules generate rich and complex structures. For most current agent-based models, however, the focus is not on the generative power of mental mechanisms as such (as it is for advocates of cultural grammar or I-culture) but on “connectionist” and “constructivist” modeling of how (micro)processes at the level of
individual decisions and actions yield macrostructural cultural norms and other social
regularities, such as spatial settlements (Dean et al, 1999), economic classes (Axtell et al, 1999),
political alliances (Axelrod & Bennett, 1993), voting patterns (Kollman et al, 1992), ecological management networks of religious water temples (Lansing & Kremer, 1993), and so on.
To the generativist, explaining the emergence of macroscopic societal regularities, such as norms or price equilibria, requires that one answer the following question: How could the decentralized local interactions of heterogeneous autonomous agents generate the given regularity? (Epstein, 1999:41)
In agent-based models of cultural phenomena there is no central, “top-down” control over individuals. Rather an initial population of autonomous heterogeneous agents, situated in a specified spatial environment, begins to interact according to rather simple local rules (e.g., if agent X manifests behavior A in the immediate, spatially proximate neighborhood of agent Y at time T, then X and Y will both manifest A at time T1; never attack an immediate neighbor; trade with a neighbor only if that neighbor is red, etc.). Over time, these concatenated individual interactions generate – or “grow” – macrostructural regularities from the “bottom up”:
Of course, there will generally be feedback from macrostructures to microstructures, as where newborn agents are conditioned by social norms or institutions that have taken shape endogenously through earlier agent interactions. In this sense, micro and macro will typically co-evolve. But as a matter of specification, no central controllers or higher authorities are posited ab initio. (Epstein 1997:42)
There is much in this approach that we find congenial, including the: 1. interpretation of society (or culture) as a dynamic and distributed computational network created by and for its constituent interacting individuals, 2. realization that individual agents have “bounded”
computing capacity and incomplete knowledge with regard to their own intentions and actions and well as the intentions and actions of others, 3. understanding that information in society is
transmitted, canalized, formed and possessed through endogenous interaction pathways (e.g., social networks), 4. realization that “emergent” macrostructural patterns and processes are neither wholly external to nor wholly internalized in individuals.7
From a cognitivist standpoint, however, the requisite mental microprocesses in current agent-based models are relatively simple (e.g., imitation, following a conventional rule, etc.).
These models also frequently incorporate functionalist views of cultural macrostructures as adaptive systems (cf. Sober, 1996) – a simplifying assumption that can lead to theoretical insights and provoke new empirical research (e.g., to the extent that cultural systems inevitably fall short of adaptive equilibrium), but which may not produce accurate descriptions or
explanations of cultural stability. This same taste for simplicity is associated with a relative neglect of ecological context (save for spatial proximity of agents) and social processes (other than dyadic contacts). These limitations are matters of practice, not principle, and reflect the goal of seeing just how much complexity can derive from minimal assumptions.
The most straightforward way to integrate our approach with agent-based modeling is to substitute empirical observations on cultural processes for the sorts of simplifying assumptions described above. Our enterprise (as well as that of other distributional theorists, such as Boyd &
Richerson, 1985) is compatible with agent-based cultural modeling. Our eventual contribution to agent-based generations of cultural macrophenomena is to: (1) enrich microspecifications of agent behaviors and decisions by specifying the cognitive mechanisms involved, and (2) furnish ethnographically plausible patterns and principles for agent behaviors and decisions. Sufficiently enriched, agent-based modeling could become a key scientific instrument for understanding the distribution and stabilization of cultural phenomena, and a potentially powerful tool for empirical research.
Summary. There are no absolute standards for evaluating different notions about what constitutes relevant cultural contents or processes and how they should be studied. Framework theories are typically judged, not by whether they are right or wrong, but rather by whether they are useful. Utility, in turn, may vary as a function of goals. All of the above approaches have strong value relative to the default condition of much of experimental psychology that focuses solely on USA undergraduates at majority universities. The relative merits of one approach versus others can be understood in terms of their positions on underlying dimensions such as scope and specificity. The situated view and cultural psychology represent two end points on this continuum. Cultural psychology aims to identify a small set of cognitive processes that (are thought to) operate very widely. Viewing cultures in terms of shared norms and values also can reveal important cultural differences. In contrast, situationists are more impressed with the lack of transfer of cognitive skills across settings (e.g., Lave and Wenger, 1991).
The framework theory that we endorse draws on insights from a number of the theories we have just reviewed. In particular our focus is on cultural processes and consequently, our approach is first cousins of both the situation and agent-based modeling approaches. We now turn to our approach and lay out its methodological and conceptual implications.
8. Cultural Epidemiology. In the norms and rules approach (including “memetics,”
Dawkins, 1976; Dennett, 1995; Blackmore, 1999; cf. Atran 2001) there is a basic assumption that memory and transmission mechanisms are reliable enough for standard Darwinian selection to operate over cultural traits (i.e., the rate of mutation is significantly lower than the selection bias). On this view, inheritable variants (of ideas, artifacts, behaviors) are copied (imitated, reproduced) with high enough fidelity so that they resemble one another more than they do
unrelated forms. Only then can they be repeatedly chosen as favorable for cultural survival or eliminated as unfavorable by selection.
We believe that these assumptions are limited because they pay insufficient attention to psychology; in particular they tend to neglect the sorts of inferential and developmental cognitive processes that allow human beings to build and participate in cultural life. For these reasons, we also believe that these various proposals for cultural “replication,” which are intended as
generalizations of Darwinian processes of replication in biology, either suffer from vagueness (e.g. memetics) or pertain to highly limited sets of phenomena (e.g. the coevolution of animal domestication and lactose tolerance in Eurasian societies, or learning by imitation). Instead, we propose to look at cultures in terms of mental representations (and attendant behaviors) that are reliably but diversely distributed across individuals in a population (the population itself being circumscribed by the intersection of these various distributions). This is what we mean by
“cultural epidemiology.”8
Boyd, Richerson and their colleagues have modeled the distributions of beliefs and practices within and across populations, and also the stabilizing role of psychological biases in transmission (Boyd & Richerson, 1985, 2001), such as conformity to preferences that already prevail in the population and emulation in deference to the beliefs and behaviors of prestigious people (Henrich & Boyd, 1998; Henrich & Gil-White, 2001). We focus on the stabilizing role of cognitive structures in the production and transmission of ideas (and attendant behaviors) that achieve widespread cultural distribution. These may not be exclusively or even mainly shared as nearly-identical mental representations across individual minds, nor transmitted more or less intact from mind to mind through any other sort of high-fidelity replication (Sperber, 1996;
Atran, 2001a). Imitation has strong limits with respect to replication - not only is it just a single
way of transmission, but also, given the many to one mappings between acts and mental representations of them (including their meaning), there’s no guarantee of any sort of fidelity.
(Indeed, “imitation” often seems to us a term of folkpsychology that needs explanation rather than explains.) We suggest that much of the cultural transmission and stabilization of ideas (artifacts and behaviors) involves the communication of poor, fragmentary and elliptical bits of information that manage to trigger rich and prior inferential structures.
The idea that cultural content may be distributionally unstable and seldom reliably
replicated is far from new (e.g. Linton, 1936, Wallace, 1961, Roberts, 1964; see Gatewood, 2001 for a review). For example, Wallace (1961, p.28) suggests that
“Culture shifts in policy from generation to generation with kaleidoscopic variety, and is characterized internally not by uniformity, but by diversity of both individuals and groups, many of whom are in continuous and overt conflict in one sub-system and in active cooperation in another.”
What may be relatively novel in our approach is the focus on variability as the object of study. The degree to which cognitive content is actually shared or similarly inferred across individual minds may depend upon many factors in addition to pre-existing cognitive structures, such as the way the physical and social environment channel the transmission of information (e.g., mountains hinder the communication and spread of ideas, classrooms facilitate them). The various distributions of ideas across populations may also be determined to a significant extent by the history of economic, political and military relations between and within groups. In the empirical body of this paper, we provide examples of how these different sorts of “canalizing”
factors – cognitive, environmental, historical – interact to produce culturally identifiable behaviors.
At the risk of some oversimplifying, we summarize the eight approaches to culture under discussion according to their stances on five key issues in Table 1. As we suggested earlier, the cultural epidemiology view is most similar to the agent-based and situated cognition views. It differs from both these views in its focus on inference. Specifically, we suggest that these pre- existing and acquired inferential structures account for the cultural recurrence and stabilization of many complexly integrated ideas and behaviors (see Boyer 1994; Atran, 2002 for religion) and set the parameters on allowable cultural diversification (Sperber & Hirschfeld, 2004).
Insert Table 1 about here
III. The Distributional View and Methodology
Our choice of methodology is chiefly motivated by theoretical concerns related to a view of cultures as variably distributed patterns of modularly constrained cognitions within given populations. Here, we describe techniques for modeling cultural consensus in order to show systematic variation in how folkbiological knowledge is put into action to generate behaviors relevant to human survival. In addition, consensus modeling permits recovery of more graded patterns of variation within and between-populations (down to the level of the individual and up to the level of combining cultural patterns to show “metacultural” interaction and consensus).
This enables us to explore the behavioral consequences and cognitive coping strategies (including patterns of information flow and exchange between individuals from different
populations) that are, for example, associated with processes of knowledge loss and devolution (described in Medin & Atran, 2004; see also Atran, Medin & Ross 2004).
It is our view that: Cultures can be effectively studied as causally distributed patterns of mental representations, their public expressions, and the resultant behaviors in given ecological contexts. People’s mental representations interact with other people’s mental representations to the extent that those representations can be physically transmitted in a public medium (language, dance, signs, artifacts, etc.). These public representations, in turn, are sequenced and channeled by
ecological features of the external environment (including the social environment) that constrain psycho-physical interactions between individuals (Sperber, 1996).
Representations that are stable over time within a culture, like those that recur across cultures, do so because they are readily produced, remembered and communicated.9 The most memorable and transmissible ideas are those most congenial to people's evolved, modular habits of mind. Arguably, some of these habits of mind evolved to capture recurrent features of hominid environments relevant to species survival (Medin & Atran, 2004). One plausible example is the apparently universal and spontaneous disposition to categorize all and only plants and animals into mutually exclusive groups of essence-based species (i.e., folk generic species, such as CAT and
REDWOOD), and to further taxonomize these groups into classes of groups under groups (e.g., folk life forms, such as MAMMAL and TREE, or folk specifics and varietals, such as TABBY and SHORT-
HAIRED TABBY). In all societies, it appears, this is done in a uniform, well-structured manner that is “not arbitrary like the grouping of stars in constellations” (Darwin, 1859:431; Berlin, 1992).
Once emitted, such core-compatible ideas will spread "contagiously" through a population of minds (Sperber, 1985). They are often learned without formal or informal teaching and, once learned, cannot be easily or wholly unlearned (Atran & Sperber, 1991). For example, when learning
about a new biological species – no matter how poor and fragmentary the stimuli (e.g., a single instance of a fuzzy drawing) - people “automatically” interpret it to have a unique causal essence that enables it to be assigned a fixed place in their folkbiological taxonomy.
A significant departure from a shared norms and rules perspective is that the variable distribution of ideas is treated as signal, not noise. The distribution view uses a set of techniques for assessing group-wide patterns that statistically demonstrate cultural consensus or lack thereof. In our work we have relied extensively on the cultural consensus model (CCM) of Romney, Weller and Batchelder (1986), an important tool for analyzing commonalties and differences within and across cultural groups. The CCM has been used as an effective tool by cognitive anthropologists (e.g. Romney and Batchelder, 1999, Romney, et al, 1996; Moore, et al, 2000).
The Cultural Consensus Model (CCM). Before describing the cultural consensus model in detail three general points are in order:1. The CCM does not prescribe which ideas should be studied any more than analysis of variance dictates which variables should be measured, 2. It is not a theory of culture but rather a tool that can be used to evaluate such theories, 3. although its most natural use and interpretation focuses on consensus, it also can be effectively used to examine within- and across-group differences (analysis of variance could also be used to detect between group differences but it is less effective in identifying within group patterns and
pinpointing cross-group differences).
The cultural consensus model assumes that widely-shared information is reflected by a high concordance among individuals. When there is a single cultural consensus, individuals may differ in their knowledge or “cultural competence.” Estimation of individual competencies is derived from the pattern of inter-informant agreement indices on the first factor of a principal
component analysis (essentially factor analysis). These competency scores should not be mistaken for scores of expertise. The cultural model provides a measure for culturally shared knowledge and hence the levels of competencies measure the extent to which an individual shares what everyone else agrees upon.
There are three standard assumptions of the original version of the CCM: 1. each item has a (culturally) “correct” answer (items are dichotomous), 2. items are independent given the culturally “correct” answers, and 3. each respondent has a fixed competence over all questions (i.e., the items are homogeneous). Batchelder and Romney have analyzed the effects of relaxing these axioms or assumptions in a number of subsequent publications (e.g. Batchelder and Romney, 1989, Romney, Batchelder and Weller, 1987; Karabatsos and Batchelder, 2003).
Although the CCM is a formal model designed for fixed-format questionnaires (true- false, matching, or multiple choice), it can also be used as a “data model” for more open-ended responses such as sorting items into a hierarchical taxonomy (see Romney, Batchelder and Weller, 1987). In this instance the individual informant data consists of an item by item matrix of distances between all pairs of items and these are used to establish similarities or agreement indices between all pairs of participants. The participant distance matrices are then correlated with each other and represent a measure of the degree to which each participant’s taxonomy agrees with every other participant’s taxonomy. The participant by participant correlation matrix is the input to the principal components analysis.
Assuming that the correlation between two informants’ sorting patterns is entirely due to the connection of each of them to the consensus knowledge, the data model approach creates a quantity in the first factor that is a proxy for consensus knowledge (Batchelder, Personal Communication, 1/26/04; see Romney, 1998 for further discussion and an application). For
interval data, the first factor loading in a principal components analysis becomes an estimate of how much an individual knows. In other words the data model provides estimates of consensus as the correlation of the individual with the aggregate. This data model is similar to reliability theory with the role of individual and item reversed, and produces an insignificant reliability overestimation compared to the formal model. This is because an item in the formal model is supposedly correlated with the cultural “truth,” whereas an individual in the data model is
correlated with an aggregate including that individual (Romney, Personal Communication, 1995;
see also Romney, et al, 1986)10.
A cultural consensus is found to the extent that the data overall conform to a single factor solution. Specifically, the first latent root - or eigenvalue - is large in relation to all other latent roots, and individual scores on the first factor are strongly positive. A good guideline is when the ratio of the first to second eigenvalue is at least 3:1.
Of course, general agreement may be coupled with systematic disagreement and the CCM is an effective tool for uncovering both shared and unshared knowledge. Another desirable characteristic of the CCM is that degree of agreement can be used to determine the minimum sample size needed to estimate the cultural consensus within some range of tolerance. In some of our studies as few as ten informants are needed to reliably establish a consensus.
Assuming that a single factor solution is satisfactory, the first factor scores for each informant represent an estimate of “cultural competence.” For example if one person has a first factor score of .90, then the best estimate is that their responses reflect 90% agreement with the overall consensus. Lower competence scores indicate less agreement with the consensus. For example, a person with a first factor score of .80 would have made more “errors” or disagree more with the consensus than the informant with a .90 score. After the consensus parameters are
estimated for each individual, the expected agreement between each pair of subjects is generated (as the product of their respective consensus parameters). In the above example the expected agreement between the two informants would be .90 times .80 (or .72). These products are used to generate an informant by informant expected agreement matrix. Next, the expected agreement matrix is subtracted from the raw agreement matrix to yield a matrix of deviations from expected agreement (cf. Hubert & Golledge, 1981). If raw and residual agreement are significantly
associated, then a significant portion of residual agreement consists of deviations from the consensus. One can then explore other factors (e.g. cultural subgroups, social network distance), which might predict or explain the residual agreement. For example, Boster (1986) found that among the Aguaruna Jívaro (Ashuar) people there was a shared cultural model for the
identification of various varieties of manioc and that deviations from this shared model were related to membership in kin and residential groups (that is, agreement within these groups is higher than what one would predict on the basis of the overall cultural model).11
Another marker for reliable residual agreement is when an analysis over two or more groups reveals systematic differences in factors beyond the first. If two groups differ in their second factor scores, then within group agreement extends beyond the overall consensus. For example, Medin et al, 1997 asked tree experts to sort local species of trees and found clear overall consensus, coupled with second factor scores correlating strongly with occupation (e.g.
parks maintenance, taxonomist, landscaper). Subsequent comparisons revealed systematic differences in the basis for sorting across these groups.
In the case of an existing consensus, the CCM justifies the aggregation of individual responses into a “cultural model.” The CCM gives an estimate of the levels of agreement among the informants. Therefore, it is possible to use this model to explore agreement patterns both
within and across different populations, the latter describing potential “meta-cultural” models.
This promotes exploration of possible pathways of learning and information exchange within and between cultural groups, illuminating more general processes of cultural formation,
transformation and evolution.
Once cultural differences are found, we can proceed to ask a series of more analytic questions about things like: 1. Are these ideas spread by means of abstract models and inference strategies or is the information conveyed in quite literal, concrete form? 2. Do factors like income or occupation or density of social networks or a variety of other input conditions
moderate cultural differences (either within or between groups)? Within the present framework the goal in studying variation is to have a theory about the distribution of ideas and flow of information, not to isolate some reified entity, "culture" (see Ross, 2004).
Summary. In a companion paper we argue that there are universal constraints on how people organize their local knowledge of biological kinds (Medin & Atran, 2004). These evolutionary constraints form a "learning landscape" that shapes the way inferences are
generalized from particular instances or experiences. It produces consensus even though specific inputs vary widely in richness and content. Thus, many different people, observing many
different exemplars of dog under varying conditions of exposure to those exemplars, may nonetheless generate more or less the same concept of dog. To say an evolved biological structure is “innate” is not to say that every important aspect of its phenotypic expression is
“genetically determined.” Biologically poised structures “canalize” development, but do not determine it – like mountains that channel scattered rain into the same mountain-valley river basin (Waddington, 1959).
In this paper, we argue that a culturally-specific learning landscape further constrains the canalization process, much as an artificially-built dam further channels the flow and shapes the path of water in a natural river basin (Sperber, 1996; Atran, 2002). The existence of any systematic distribution of ideas and behaviors, or cultural “path,” results from an integration of distinct cognitive, behavioral, and ecological constraints that neither reside wholly within minds nor is recognizable in a world without minds (in this respect we agree with the situated cognition view). Cultural paths do not exist apart from the individual minds that constitute them and the environments that constrain them, any more than a physical path exists apart from the organisms that tread it and the surrounding ecology that restricts its location and course. As Inagaki and Hatano (2002) suggest, it is the confluence of these various sources of constraints that makes cross-cultural comparisons possible and meaningful.
IV. Garden Experiments in Mesoamerica and North America
There is little or no detail available in typical normative accounts of social structure in the anthropological literature that would allow evaluation of patterns of individual variation,
agreement and disagreement within and between groups. Without such detail, normative claims are difficult to verify or falsify. The overarching reason is simple: anthropologists are typically instructed to go out into the field alone for some months or – in exceptional cases – some few years and bring back a description of the society studied. The popular image of the
anthropologist with a pith helmet and notebook is not very far off the mark; only now the pith helmet is a baseball cap or canvas fedora, and the notebook is a PC. In this situation, there is little alternative to normative description (excepting the “narratives” of anti-positivist post- modernism, which do little to foster dialogue with the larger scientific community).
Detailed analyses of the relations between ecology, technology, social networks and so forth require large interdisciplinary efforts, over many field seasons, at a cost that usually exceeds typical ethnographic fieldwork by one or more orders of magnitude. The pertinent academic and government funding institutions are not set up for this kind of project, and so the effort is rarely made (for a notable exception, see Henrich, et al, 2001). We have been fortunate to be involved in two such efforts: one in Mesoamerica and another in North America.
A case study on the importance of cultural models versus environmental determination comes from a variation on the “common garden experiment” in biology. When members of a species have different phenotypes in different environments, samples are taken from both environments and replanted in only one. If the differences still exist, they are probably genetic (two genotypes); if not, then they are probably environmental (one genotype producing two phenotypes). Here we use a variation on this experimental approach. Our aim is not to distinguish genetic nature from environmental nurture but rather to isolate the role of certain socio-cultural factors (social networks, cognitive models) from other economic (sources and level of income), demographic (family and population size), and ecological factors (habitat and species) in environmental management. Evidence for the importance of culturally transmitted factors on behavior would be data showing that groups of people who have different cultural histories and cultural ideas behave differently in the same physical environment.
Study Populations and Context. In the next several paragraphs we describe the main study populations in our research.
Mesoamerican populations. A key focus of our work concerns three cultural groups in the same municipality in Guatemala’s Department of El Petén: native Itza' Maya, Spanish- speaking immigrant Ladinos and immigrant Q’eqchi’ Maya. Itza’ and Q’eqchi were each
circumscribed by entirely overlapping and perfectly redundant criteria of proximity of residence, ethnic self-identification, and a multigenerational history of pervasive family interconnections.
The Q’echi’ were also identified by their mother tongue. The Ladino population was initially circumscribed by proximity of residence, language (Spanish), ethnic self-identification and lack of a community-wide history of family interconnections. In addition, members of each
community readily and distinctly identify the group affiliation of members of the other
communities. This initial circumscription of cultural groups obviously relied on commonsense conceptions of cultural differences, but our analyses and subsequent findings were not bound by these initial selection criteria (see the observations on circularity in cultural research in the General Discussion).
In all groups, men are primarily occupied with practicing agriculture and horticulture, hunting game and fish, and extracting timber and non-timber forest products for sale. Women mainly attend to household gardening and maintenance. The climate is semitropical, with quasi- rainforest predominating (tropical dry forest or hot subtropical humid forest). Topographic and microclimatic variation allow for a dramatic range of vegetation types over relatively small areas, and sustaining both this diversity and people’s livelihood over the last two millennia has required correspondingly flexible agro-forestry regimes (Sabloff & Henderson, 1993; Atran, Lois
& Ucan Ek’, 2004).
Native Itza’ Maya. The Itza’, who ruled the last independent Maya polity, were reduced to corvée labor after their conquest in 1697 (Atran, et al, 2004). San José was founded as one of a handful of "reductions" for concentrating remnants of the native Itza’ population (and fragments of related groups). In 1960, the military government opened the Petén (which includes 35,000 km2, about 1/3 of Guatemala's territory) to immigration and colonization. In the following years,
about half the forest cover of Petén was cleared. In a project engineered by the Agency for International Development (USAID) and supported by a debt for nature swap, Guatemala’s government set aside remaining forests north of 17˚10' latitude as a Maya Biosphere Reserve, a designation recognized by UNESCO in 1990. The San José municipality now lies within the Reserve’s official “buffer zone” between that latitude and Lake Petén Itza’ to the south. Today San José has some 1800 habitants, about half of whom identify themselves as Itza’, although only older adults speak the native tongue (a Lowland Mayan language related to Yukatek, Mopan and Lakantun).
Immigrant Ladinos. The neighboring settlement of La Nueva San José was established in 1978 under jurisdiction of the Municipality of San José. The vast majority of households (about 600 people) are Ladinos (native Spanish-speakers, mainly of mixed European and Amerindian descent) most of whom were born outside of Petén. The majority migrated to the area in the 1970s as nuclear families stemming from various towns of southern Guatemala.
Q’eqchi’ Maya .The hamlet of Corozal, also within the Municipality of San José, was settled at the same time by Q’eqchi’ speakers, a Highland Maya group from the Department of Alta Verapaz just south of Petén. Q’eqchi’ filtered in as nuclear families, migrating in two waves that transplanted partial Highland communities to Corozal: a) directly from towns in the vicinity of Cobán (capital of Alta Verapaz), b) indirectly from Alta Verapaz via the southern Petén town of San Luis (home to a mixed community of Q’eqchi’ and Mopan Maya). Q’eqchi’ immigration into Petén began as early as the 18th century, though massive population displacement into Petén is recent. The Q’eqchi’ now constitute the largest identifiable ethnic group in Petén while
maintaining the smallest number of dialects and largest percentage of monolinguals (Wilson 1995:38; cf. Stewart 1980). This reflects the suddenness, magnitude, and relative isolation of the
Q’eqchi’ migration. Although many of the nearly 400 Q’eqchi’ of Corozal understand Spanish, few willingly converse in it. Q’eqchi’ is not mutually intelligible with Itza’. To help understand results with Lowland Q’eqchi’ immigrants, we also studied a native Highland Q’eqchi’ group.
North American Populations. Resource conservation and conflict over it are also important issues in Wisconsin where majority culture and Native American hunters and fishermen may conceptualize nature in distinct ways.
Menominee. The Menominee (“Wild Rice People”) are the oldest continuous residents of Wisconsin. Historically, their lands covered much of Wisconsin but were reduced, treaty by treaty, until the present 95,000 hectares (about 240,000 acres) was reached in 1854. There are 4- 5000 Menominee living on tribal lands in and around three small communities. Over 60% of Menominee adults have at least a high school education and 15% have had some college. The present site was forested then and now - there are currently about 88,000 hectares of forest.
Many of the vast Great Lakes forests did not survive the post civil war flurry of logging. In contrast, Menominee have practiced logging on a sustainable basis for the last 150 years. The Menominee reservation is managed by a tribal legislature. Sustainable coexistence with nature is a strong value (Hall & Pecore, 1995). Hunting and fishing are important activities for most adult males and for many females. Fishing is practiced on the reservation’s many lakes and streams.
The tribe sets specific rules for both fishing and hunting on the reservation, one of them specifically outlawing the “wanton destruction” of any species.
Rural Majority Culture. Adjacent to the Menominee reservation is Shawano County, which consists of farmland, small forests, and numerous lakes and rivers. The county was well- settled by 1850 and a substantial portion of the current population can trace its origins in the area to 19th century immigration. As for the Menominee, hunting and fishing plays a big role in the
social and recreational life of the Majority Culture people. Most fishing is on the Wolf River or Shawano Lake (which also attracts a lot of fishing tourists). Compared to Menominee fishermen, fishing for food source is a lesser priority for majority-culture fishermen, who are more
interested in fishing for sport. Thus, it is not surprising that catch and release fishing plays a somewhat more important role off the reservation than on it. Finally, while Menominee can purchase a fishing license for off-reservation fishing, majority culture people cannot fish on the reservation. Fishing is done during all seasons, including ice-fishing in winter.
V. Folkecology and the Spirit of the Commons.
In earlier studies, we found that Itza’ Maya informants consistently appealed to
ecological relations on category-based induction tasks unlike USA college students who focused on taxonomic relations: for example, generalizing a property of bats to horses on the grounds that horses are likely to be bitten by bats (reviewed in Medin & Atran, 2004). Although we did not have corresponding data from Q’eqchi’ and Ladino adults on such tasks, the Itza’ propensity for ecological reasoning, coupled with their record of sustainable agro-forestry, suggested to us that there may be a connection between folkecological models and behavior. Furthermore, the fact that the Ladino and Q’eqchi’ populations practice agro-forestry in a much less sustainable manner, provided the opportunity to explore whether understandings of the forest are correlated with action on it. These conjectures led us to a series of systematic cross cultural and within cultural comparisons that are pertinent to a variety of conceptual issues in cognition, decision making, and culture theory (Atran et al.,1999a, 2002; Medin et al., 2002; see also Ross 2002;
Ross & Medin, in press). Evidence for the importance of culturally transmitted cognitive models on behavior would indicate that groups of people who have different cultural histories and cultural ideas behave differently in the same physical environment.
We have employed a threefold approach to understanding causal relations between individual cognitions, human behaviors that directly affect the environment, and cultural patterns that emerge from population-wide distributions of cognitions:
1. Folkecology. This involves a cross-cultural methodology for modeling people's cognitions of the ecological relationships between plants, animals and humans.
2. Cultural Epidemiology. This involves ways of mapping individual variation and inter- informant agreement in the flow of ecologically-relevant information within and between societies, using social network analysis to trace potential transmission pathways in transfer of knowledge.
3. Spiritual Values and the Commons. This involves operationalizing the role of “non- economic” entities and values, such as supernatural beings, in environmental cognition and behavior.
The Lowland Maya region faces environmental disaster, owing in part to a host of non- native actors having access to the forest resources (Schwartz, 1995). A central problem concerns differential use of common-pool resources, such as forest plants, by different cultural groups exploiting the same habitat. As we noted earlier, one strong view is that individual calculations of rational self-interest collectively lead to a breakdown of a society’s common resource base unless institutional mechanisms restrict access to co-operators (Hardin, 1968, Berkes et al., 1989). The reason is clear: in the absence of monitoring and punishment, exploiters gain the same benefits as co-operators but at reduced cost. Co-operators are driven to extinction, and exploiters flourish until the commons is destroyed. Still, exclusive concern with economic rationality and institutional constraints on action may not sufficiently account for differences in
